Local Excitation of the 5-Bromouracil Chromophore in DNA

ARTICLE pubs.acs.org/JPCB

Local Excitation of the 5-Bromouracil Chromophore in DNA. Computational and UV Spectroscopic Studies Piotr Storoniak,*,† Janusz Rak,† Katarzyna Polska,† and Lluís Blancafort‡ † ‡

Department of Chemistry, University of Gda
nsk, Sobieskiego 18, 80-952 Gda
nsk, Poland Institute of Computational Chemistry, Department of Chemistry, University of Girona, Campus de Montilivi, 17071 Girona, Spain

bS Supporting Information ABSTRACT: The UV electronic transition energies and their oscillator strengths for two stacked dimers having B-DNA geometries and consisting of 5-bromouracil (BrU) and a purine base were studied at the MS-CASPT2/6-311G(d) level with an active space of 12 orbitals and 12 electrons. The calculated energy of the first vertical (π,π*) transitions for the studied dimers remain in fair agreement with the maxima in the difference spectra measured for duplexes with the 50 -ABrU-30 or 50 -GBr U-30 sequences. Our MS-CASPT2 results show that the charge transfer (CT) states in which an electron is transferred from A/G to BrU are located at much higher energies than the first (π,π*) transitions, which involve local excitation (LE) of BrU. Moreover, CT transitions are characterized by small oscillator strengths, which implies that they could not be excited directly. The results of the current studies suggest that the formation of the reactive uracil-5-yl radical in DNA is preceded by the formation of the highly oxidative LE state of BrU, which is followed by electron transfer, presumably from guanine.

1. INTRODUCTION Halogen-substituted pyrimidines have been investigated since the 1960s owing to their potential application in the photo therapy of cancer diseases:1 it was demonstrated that halogenouracils, notably 5-bromouracil (BrU), when incorporated in DNA, sensitize it to UV radiation. Indeed, exposure of DNA containing 5-bromouridine to radiation of 250
320 nm produces several types of DNA damage such as direct strand breaks,2
4 alkali-labile sites,5
8 and intra-9
11 and interstrand12 photocross-links. These effects are attributed to a highly reactive vinyl uracil-5-yl radical, produced from BrU by UV quanta.5,13 The damaging action of the uracil-5-yl radical seems to be related to its ability to abstract hydrogen atoms from the sugar constituents of DNA. Early reports presumed that a single strand break occurring in the vicinity of 5-bromouracil could be brought about when an H atom was abstracted from the 50 -adjacent sugar by the uracil-5-yl radical formed during photolysis.5,13 This supposition remains in accordance with the well-known fact that the loss of an H atom from any position of the deoxyribose moiety may lead to a direct strand break in DNA.14 The first detailed proposal of the damage mechanism was put forward by Saito and Sugiyama.8 Using UV light of 302 nm in aqueous solutions containing short duplexes labeled with BrU, these authors demonstrated significant damage enhancement for the 50 -ABrU-30 sequence as compared to duplexes containing the 50 -GBrU-30 fragment. To rationalize this sequence-selectivity, it was originally postulated that photochemical decomposition of Br U is initiated by photoinduced single electron transfer (PSET) r 2011 American Chemical Society

from a 50 -neighbor adenine, followed by the immediate elimination of a bromide anion from the resulting BrU•
, which leads to the formation of the uracil-5-yl radical. The latter species can abstract hydrogen from the C10 position of the deoxyribose of the adjacent nucleobase. The radical centered on the sugar is then oxidized by the cation radical of a nucleobase (the second product of PSET is a nucleobase radical cation). Finally, the formation of a carbocation on C10 is followed by water-assisted depurination. In subsequent reports hydrogen atom abstraction from C20 was also considered.15 This mechanism was later adapted by Greenberg and coworkers2 to explain UV-induced strand break formation in the models of B-DNA labeled with BrU. In their photolytic experiments, carried out under anaerobic conditions, strand scissions at the 50 -ABrU-30 sequence were generated approximately 8 times more efficiently than those at the 50 -GBrU-30 one.2 Since guanine has a substantially lower ionization potential than adenine,16 this process has been dubbed “contrathermodynamic”. In order to explain the experimental picture it was assumed that fast back electron transfer in the 50 -GBrU radical ion pair prevents the loss of a bromide ion from BrU and therefore suppresses strand damage.2,8,17 The dependence of photosensitization efficiency on the wavelength and conformation of DNA labeled with BrU was studied by Received: January 31, 2011 Revised: March 11, 2011 Published: March 28, 2011 4532

dx.doi.org/10.1021/jp201028a | J. Phys. Chem. B 2011, 115, 4532–4537

The Journal of Physical Chemistry B Cecchini et al.12 They demonstrated that UV irradiation efficiently induces single strand breaks at the BrU site in a completely cDNA duplex (no strand breaks were observed in the complementary nonbrominated strand). On the other hand, the efficiency of single strand cleavage in the duplex containing a bulge at the BrU site was much lower, and cleavages were observed at both strands in the open DNA region. Moreover, for such a duplex the main photoproducts turned out to be interstrand cross-links.12 The increased efficiency of UV-induced strand break damage in double-stranded BrU-labeled DNA was also demonstrated by other authors.3,18 A decrease in strand break formation in nonhybridized oligonucleotides was explained by the decreased π-stacking in a single-stranded DNA and therefore less efficient PSET.18 Very recently, the 50 -GAABrU-30 sequence rather than the 50 Br A U-30 one was proposed as a “hotspot”, whose presence in double-stranded DNA guarantees the high photosensitivity of the biopolymer.19 The results of several UV experiments with duplexes of different sequences demonstrated convincingly that longrange electron transfer from guanine to BrU (or 5-iodouridine) is responsible for the observed photoreactivity.19
21 Indeed, the maximum damage was observed when two AT pairs were present in the bridge connecting distant G and BrU, whereas photoreactivity was lowered with both the shortening and lengthening of the AT bridge.19,20 Moreover, substitution of G at the hotspot with inosine or adenine, which have substantially larger ionization potentials than guanine, made these duplexes completely nonphotoreactive.19 Although the long-range electron transfer induced by photoexcitation of BrU in DNA seems to be well-founded,19
21 the primary process responsible for the damage is still ambiguous. In particular, it is unclear whether the electronic excitation of BrU leads instantaneously to charge separation, or whether BrU is first locally excited, forming a state with a high oxidation potential, which is only then followed by electron transfer from a distant nucleobase. Using difference UV spectroscopy, we show in the current work that the substitution of uracil by BrU in two DNA duplexes, one containing a 50 -UA-30 fragment and the other a 50 -UG-30 fragment, leads to absorption peaking around 290 nm. Using MS-CASPT2 ab initio calculations, BrU is identified as the absorbing species around this wavelength, i.e., the excitation is initially localized in the BrU molecule. The corresponding charge transfer states, involving BrU and A or BrU and G, are separated from the locally excited state by 1.22 and 0.53 eV (28 and 12 kcal/mol) and should produce a blue-shifted difference spectrum with respect to the measured one. The results of our computational
experimental studies suggest that electronic excitation of labeled DNA with 300 nm photons leads to locally excited BrU, which may subsequently undergo long-distance charge transfer, leading to the damage observed.

2. METHODS 2.1. Computational Details. Since literature reports suggest that a 50 -neighboring base is crucial for the photoreactivity of BrU in DNA, we chose to model the initial absorption event in the gas phase with two DNA stacks consisting of purine and bromouracil: 50 -ABrU-30 and 50 -GBrU-30 (see Figure 1). The geometries of those stacks were generated with the help of the X3DNA22 program, using averaged X-ray and neutron crystal geometries of nucleobases,23 and the distance between adjacent bases of 3.38 Å.

ARTICLE

Figure 1. Stacked dimers, ABrU and GBrU (50 f30 direction), used as molecular models.

The CASSCF (complete active space self-consistent field) reference functions for the stacked bases were used for calculations at the CASPT2 (complete active space second-order perturbation) level to account for dynamic correlation (denoted by CASPT2/ CASSCF). In order to further refine the energetics the multistate extension of CASPT2 (denoted by MS-CASPT2/CASSCF) was applied. The calculations were carried out with the triple-ζ 6-311G(d) basis set, which represents a compromise between computational cost and accuracy for the description of the valence excitations. Calculations for isolated adenine (see Results) are in good agreement with previous CASPT2/6-31G(d) and CASPT2/6-311þG(d,p) results.24 Diffuse functions were not included in the basis set to prevent the appearance of Rydberg states. In the monomers, these states mix with the valence states of the same symmetry and have a substantial effect on the oscillator strengths.25 For the G and A monomers, they appear near the range of energies of interest to us (approximately 5.0 and 5.3 eV, respectively). However, one can assume that the energy of these states increases upon Watson
Crick pairing because the diffuse orbitals are mainly near the NH2 groups involved in the pairing. Therefore, they can be safely neglected for our purposes. Gaussian0326 was employed for the CASSCF/6-311G(d) calculations, and the electronic energies were recalculated at the CASPT2 and MS-CASPT2 levels using MOLCAS-6.4.27 CASSCF Active Spaces. The complete valence active space of the ABrU and GBrU dimers consists of 32 electrons in 23 orbitals, and 34 electrons in 24 orbitals, respectively. This active space can safely be reduced by excluding the oxygen and nitrogen lone pair orbitals. Thus, the (n,π*) transitions will play a minor role in the absorption because of their small oscillator strengths. Moreover, hydrogen bonds due to Watson
Crick pairing will raise the energy of these states well above the experimental excitation wavelength of 300 nm. The remaining active space contains all π orbitals and consists of 22 electrons in 18 orbitals for ABrU, i.e., (22,18), and (24,19) for GBrU. This is still well beyond the current computational capabilities, and therefore the active space was further reduced to (12,12), where every base contributes six electrons and six orbitals to the active space. To calibrate the (12,12) active space for the stacked base pair, the MS-CASPT2/CASSCF/6-311G(d) excitation energies of the (π,π*) states of the isolated bases were calculated with a (6,6) active space and compared with the energies calculated with the full π active space, i.e. (10,8) for BrU, (12,10) for adenine and (14,11) for guanine. CASPT2 and MS-CASPT2 Calculations. The CASPT2 and MS-CASPT2 energies of ABrU and GBrU dimers were calculated with the (12,12) active space. The calculations were done using a state-averaged seven-root CASSCF reference function with equal weights for all states. Such a large number of states is necessary 4533

dx.doi.org/10.1021/jp201028a |J. Phys. Chem. B 2011, 115, 4532–4537

The Journal of Physical Chemistry B

ARTICLE

Table 1. MS-CASPT2/CASSCF Vertical Excitation Energies (Eex) and Oscillator Strengths (f) for the Isolated 5-Bromouracil, Adenine, and Guanine Monomers

Table 2. Vertical Excitation Energies (Eex), Oscillator Strengths (f), and Dipole Moment Differences in the y Direction (Δμ(Y)) for Adenine-Bromouracil (ABrU), Calculated at the MS-CASPT2(12,12)/6-311G(d) Level

5-bromouracil (10,8)/6-311G(d) Eex (eV)

(6,6)/6-311G(d)

λ (nm)

a

f

Eex (eV)

λ (nm)

fa

S1

4.73

262

0.261

4.76

260

0.322

S2

5.78

214

0.050

6.15

201

0.061

adenine (12,10)/6-311G(d) Eex (eV)

λ (nm)

f

Eex (eV)

λ (nm)

fa

S1

4.80

258

0.053

4.97

249

0.008

S2

4.99

249

0.214

5.51

225

0.169

guanine (14,11)/6-311G(d)

(6,6)/6-311G(d)

Eex (eV)

λ (nm)

f

Eex (eV)

λ (nm)

fa

S1

4.78

259

0.146

4.69

264

0.306

S2

5.49

226

0.098

5.69

218

0.133

a

λ (nm)

fa

Δμ(Y)b

S1

4.56

272

0.246

0.004

LEBrU

S2

5.18

239

0.006

0.197

LEA

S3

5.52

224

0.163

0.110

LEA

S4

5.78

215

0.003

6.088

CTAfBrU

S5

6.11

203

0.069

0.216

LEBrU

S6

6.88

180

0.003

6.515

CTAfBrU

statec

a

(6,6)/6-311G(d) a

Eex (eV)

a

Oscillator strength calculated from MS-CASPT2 energy differences and PM-CASCI transition dipole moments.

because the charge transfer states appear as sixth or seventh root of the CASSCF calculation. The MS-CASPT2 calculations were carried out with no ionization potential electron affinity (IPEA) correction,28 and a real level shift parameter of 0.3 au29 was used for the dimers. For the isolated monomers, the real level shift was 0.2 au (full π active space calculations) and 0.4 au ((6,6) calculations). The oscillator strengths are obtained using the transition dipole moments of the perturbationally modified complete active space configuration interaction (PM-CASCI) wave function and the MS-CASPT2 energies. 2.2. Experimental Details. Absorption Spectra. In order to compare the absorption characteristic of the 50 -ABrU-30 and 50 -GBrU-30 sequences in DNA we registered the UV spectra of two double-stranded sequences: 50 -ATA TGC GXG CTA GCG-30 and 50 -ATA TGC AXG CTA GCG-30 , where X = BrU. The HPLC grade oligonucleotides were dissolved in 0.5 mM Tris buffer (pH = 8). The spectra were recorded with a Nanodrop ND-1000 spectrophotometer for 2 μL samples of concentration around 420 ng/μL. The spectra of the unsubstituted, natural oligomers (X = U) were recorded under the same conditions to obtain the difference spectra used for the assignments.

3. RESULTS 3.1. Active Space Calibration. Table 1 shows the vertical excitation energies for the isolated 5-bromouracil, adenine, and guanine bases, calculated at the MS-CASPT2 level of theory with the (6,6) active and full π active spaces. Since we have used our calculations to assign the absorption spectrum around 300 nm, our aim is to reproduce with reasonable accuracy the excitation energies and oscillator strengths of the stacked dimer states that

Oscillator strength calculated from MS-CASPT2 energy differences and PM-CASCI transition dipole moments. b PM-CASCI dipole moment differences. c States characterized by the PM-CASCI wave function.

come around or below 5 eV. In this respect, there is satisfactory agreement between the (6,6) and full π active space calculations. For 5-bromouracil, the lowest (π,π*) state appears at 4.76 eV with the (6,6) active space (see Table 1). This value is in good agreement with the value of 4.73 eV obtained with the larger active space and the experimental value of 4.66 eV obtained with electron energy loss (EEL) spectroscopy.30 There is a discrepancy of approximately 0.3 eV with respect to the value calculated by some of us in a previous 5-bromouracil study, but this is probably due to the different geometries used in the present study (idealized experimental structure) and the previous one (structure optimized at the CASSCF level).31 For S2 we observe larger discrepancies between the (6,6) and (10,8) values, but this should only be of minor importance in this study because we are interested in the low energy excited states. Turning to the purine bases, in Table 1 we show the energies for the two lowest (π,π*) transitions of 1La and 1Lb character. For adenine, both states appear at 4.97 eV (1Lb state with small oscillator strength) and 5.51 eV (bright 1La state) with the (6,6) active space. The value of the 1Lb state is in good agreement with the (12,10) value (4.80 eV). However, the agreement is less satisfactory for the 1La state, which is blue-shifted by approximately 0.5 eV with respect to the (12,10) value (4.99 eV) and the available experimental estimations available (approximately 4.9 eV).32 For guanine, the agreement between the (6,6) and the (14,11) values is good for both states. The 1La state appears at 4.69 and 4.78 eV, respectively. Taking into account the fact that the 0
0 transition for the isolated 9-H keto form of guanine was measured at 4.2 eV with the help of supersonic jet experiments33 and supplemented with the energy difference between the vertical and adiabatic transitions in this guanine tautomer, as calculated by Marian at the multireference configuration interaction
density functional theory (MRCI-DFT) level,34 one can estimate that the vertical excitation of gaseous guanine should be observed at ca. 4.72 eV. There is also good agreement for the calculated 1Lb excitation energy, which is 5.69 and 5.49 eV with the (6,6) and (14,11) active spaces. The calculated vertical excitations are also in good agreement with the previous CASPT2 values obtained by Serrano-Andr
es and co-workers with a larger ANO-L basis set.35 For the three bases studied, the oscillator strengths are also described satisfactorily with the (6,6) active space as compared to the data with the full π active spaces (presented in Table 1). The 4534

dx.doi.org/10.1021/jp201028a |J. Phys. Chem. B 2011, 115, 4532–4537

The Journal of Physical Chemistry B

ARTICLE

Table 3. Vertical Excitation Energies (Eex), Oscillator Strengths (f), and Dipole Moment Differences in the y Direction (Δμ(Y)) for Guanine-Bromouracil (GBrU), Calculated at the MS-CASPT2(12,12)/6-311G(d) Level Eex (eV)

λ (nm)

fa

Δμ(Y)b

S1

4.37

283

0.049

0.367

LEBrU

S2

4.71

263

0.519

1.067

LEG

S3

4.90

253

0.071

5.491

CTGfBrU

S4

5.88

211

0.080

0.038

LEG

S5

6.30

197

0.029

0.162

LEBrU

S6

7.58

163

0.048

CTGfBrU/CTBrUfG

10.
7

statec

a

Oscillator strength calculated from MS-CASPT2 energy differences and PM-CASCI transition dipole moments. b PM-CASCI dipole moment differences. c States characterized by the PM-CASCI wave function.

largest deviation is found for the 1Lb state of adenine, where there is a deviation by a factor of approximately 6. From our analysis we can conclude that the error caused by reducing the active space is about 0.1 eV for the excitation energies of low energy excitations, while the oscillator strengths are calculated within the same order of magnitude as with the full π active space. 3.2. Vertical Excitation of Dimers. Based on the ability of the (6,6) reference active spaces to describe the lowest excited states of the nucleobases of interest, the calculations for the stacked dimers were carried out with an active space consisting of 12 electrons and 12 orbitals distributed evenly between the two bases. The MS-CASPT2 vertical (π,π*) transition energies, oscillator strengths, and dipole moment components obtained with this approach are listed in Tables 2 and 3. The CASSCF and CASPT2 transition energies as well as the orbitals used to construct the active space are presented in Tables S4 and S5 and Figures S1 and S2 in the Supporting Information. Each state is characterized as locally excited or charge transfer on the basis of the semioccupied orbitals of the dominant PM-CASCI configurations. The charge transfer character of the states is monitored by further calculating the differences in the dipole moment along the stacking direction (y axis in our case) between the ground state and the state of interest. For a stacking distance of 3.38 Å, a pure charge transfer state should have a dipole moment difference of 6.76 au, while for the locally excited states the difference should be 0 au. Small deviations from these values may be due to the small charge transfer character of the ground state; in the states with larger deviations (e.g., S1, S2, and S3 of the GBrU dimer) the states have a mixed character due to the mixing of the close lying charge transfer and locally excited states. For both dimers, the lowest energy transition corresponds to the local excitation of BrU (see Tables 2 and 3). These transitions are red-shifted with respect to that in the isolated 5-bromouracil shown in Table 1. Indeed, the MS-CASPT2 value of the excitation energy for gaseous BrU amounts to 260 nm, whereas for the ABrU and GBrU dimers it is 272 and 283 nm, respectively. These data suggest that guanine induces a stronger red shift of the electronic transition in BrU than adenine. Moreover, the oscillator strengths connected with the excitation of BrU in the dimers are smaller (0.246 in ABrU and 0.049 in GBrU; see Table 2 and 3) than in the monomer (0.322) (see Table 1). The decrease in the oscillator strength is due to coupling with charge transfer states, which have much smaller oscillator strengths. For instance, orbital analysis in GBrU shows that the S1 excitation is localized mainly in the 5-bromouracil base, but the dipole

Figure 2. Difference spectra for duplexes containing the ABrU and GBrU sequences.

moment difference with respect to the ground state indicates that the charge transfer state contributes about 5% to the calculated state. In turn, the transition dipole moment for the charge transfer state of GBrU increases owing to intensity borrowing from the local excitations. The first charge transfer state (where an electron is transferred from purine to 5-bromouracil) was calculated at 215 nm (the S4 state; see Table 2) for the ABrU stack and at 253 nm (the S3 state; see Table 3) for the GBrU one. Note that the relative positions of those charge transfer states follow the trend exhibited by the ionization potentials of adenine and guanine, i.e., the transition was calculated at a higher energy for ABrU. In GBrU, the charge transfer state has a larger oscillator strength than that for ABrU due to intensity borrowing from the nearby locally excited states, as explained above. The CT states are separated from the lowest energy locally excited states by 0.53 and 1.22 eV for GBrU and ABrU, respectively (see Table 2 and Table 3). These values, together with the calculated oscillator strengths, indicate that the direct CT process should be favored in GBrU as compared to ABrU, since the oscillator strengths for the latter are ca. 24 times lower than that for GBrU. This indirectly supports our conclusion that the CT states are not populated directly during the excitation, since the yield of oxidation product in duplexes containing the GBrU and ABrU sequences follows the opposite trend.4 3.3. Experimental Difference Spectra. The difference spectra for two pairs of duplexes containing the 50 -AX-30 and 50 -GX-30 sequences (X = U and BrdU) are shown in Figure 2. This procedure allows one to extract the UV absorption originating during the substitution of uracil by 5-bromouracil from the total absorption of the duplex. For the duplex containing the ABrU sequence, the maximum of the difference spectrum is centered around 285 nm, while for the duplex containing the GBrU sequence the maximum is red-shifted to 290 nm. The band extends up to 260 nm, and no additional absorption is observed at higher energies.

4. DISCUSSION The difference spectra show that the substitution of U by BrU in the duplexes leads to the appearance of a band centered at 290 nm for the GBrU-containing duplex, and 285 nm for the ABrU-containing one (Figure 2). Our calculations suggest that it corresponds to a local excitation of the 5-bromouracil chromophore rather than to a charge transfer state from the neighboring purine bases. Moreover, the red shift of the BrU absorption in the duplexes with respect to the vertical excitation of the isolated base is due to π-stacking. In our calculations we have only 4535

dx.doi.org/10.1021/jp201028a |J. Phys. Chem. B 2011, 115, 4532–4537

The Journal of Physical Chemistry B considered stacking with the purine base lying at the 50 end, but this seems to be the major cause of the red shift since we have been able to reproduce the experimental maxima. The direct excitation of the charge transfer state during the excitation is the hypothesis favored in the literature, but our calculations suggest that it can be safely ruled out since the CT states appear at substantially higher energies and have lower oscillator strengths. Indeed, the CT state for the GBrU and ABrU duplexes are 0.5 and 1.2 eV higher than the local BrU excitation energies (see Tables 2 and 3). These values are gas-phase estimates, but inclusion of environmental effects will not alter the global picture because the DNA is only moderately polar. The stabilization of the CT states can be estimated with the Bell model.36 Using a value of 4.8 for the dielectric constant (see, for instance, the modeling of electron transfer between rhodium(III) complexes and the pyrimidine photodimer in DNA)37 and cavity radius 5 Å, the estimated stabilization of the CT state with respect to the locally excited states is 0.06 and 0.03 eV for ABrU and GBrU, respectively. Even if this shift is included in the calculation, the excitation energy of the CT states lies only at the tail of our measured difference spectra. The experimental maxima of the difference spectra are close to the 300 nm wavelength used in the photochemical experiments on 5-bromouridine-labeled DNA fragments.2
4,8,12,17,19 Our analysis shows that the primarily excited species in these experiments is the 5-bromouracil chromophore with a localized excitation. The subsequent formation of the CT state is due to the high oxidative potential of this state. Thus, on the basis of the MS-CASPT2 vertical excitations and the adiabatic electron affinity of 5-bromouracil in its ground state (0.63 eV),38 we can provide an upper bound of 5.0 and 5.2 eV, respectively, for the electron affinity of the excited state of 5-bromouracil in the GBrUand ABrU-containing strands. The actual values may be lower because of the relaxation of the primary excited species, but in any case the local excited state will be a strong oxidant. Consequently, from our present results we propose a three-step mechanism for photodamage in DNA sensitized with BrU: (i) Photons of 300 nm excite BrU to the first excited state and generate a highly oxidative species. (ii) The locally excited BrU chromophore oxidizes a nearby guanine base and generates the charge separated species. (iii) The generated radical anion of 5-bromouracil releases a bromine anion, and the resulting uracil5-yl radical triggers the chemical transformations that lead to the damage. This mechanism is an alternative to the one usually accepted in the literature, where the reactive BrU•
species is formed directly during the excitation through photoinduced single electron transfer. Note that a stepwise mechanism for CT state formation analogous to the one described here has been postulated for dinucleotide monophosphates in solution.39 It should be clear that this mechanism is embedded in a complex kinetic scenario, where it competes with other processes. For instance, the second step of our mechanism (formation of the charge-separated species from locally excited BrU) may compete with other decay mechanisms of the locally excited species. One of them is internal conversion to the (n,π*) state, but this process is unlikely in the DNA environment because the energy of this state will be increased because of the hydrogen bonding of the carbonyl oxygen in Watson
Crick pairing. Another possible competing mechanism is the dissociation of the neutral, excited chromophore, which has been observed experimentally for BrU in solution and has been described in our recent theoretical study on the photodissociation of

ARTICLE

isolated 5-bromouracil.31 However, the occurrence of this process in DNA seems unlikely because the photodamage is not observed in modified duplexes containing the 50 -BrUT-30 sequence.19 The third competing mechanism is the benign, radiationless decay of the locally excited species to the ground state through a seam of conical intersection. This mechanism has also been described in our computational study and will be the main competing mechanism with the formation of the charge transfer species.31 Finally, the third step of our mechanism (dissociation of the BrU radical anion) will compete with back electron transfer to the guanine base. One of the points of controversy with respect to the mechanism of photoinduced damage in modified DNA relates to the greater damage incurred by fragments containing the 50 -ABrU-30 sequence compared to those containing the 50 -GBrU-30 one. Indeed, G has a lower oxidation potential than A, and one may expect to find more damage in the 50 -GBrU-30 -containing duplexes. Recently, Sugiyama et al. partially solved this controversy, showing that the oxidized species is always a guanine base.19 Thus, in the 50 -ABrU-30 -containing fragments the damage is produced through long-range electron transfer from guanine at a different position. In this context, our calculations suggest that the higher efficiency experimentally found for the duplexes containing the 50 -ABrU-30 sequence comes from the greater intensity of the localized BrU excitation in these fragments compared to the duplexes containing the 50 -GBrU-30 sequence. However, as we have discussed above, the lowering of the oscillator strength in the 50 -GBrU-30 stack is due to the interplay between the lowest excited states of the stack, which may be substantially dependent on conformational and environmental effects; more sophisticated calculations will be necessary to confirm this suggestion.

5. SUMMARY The spectral characteristics of two stacked dimers having B-DNA geometries and consisting of BrU and a purine base were studied at the MS-CASPT2/6-311G(d) level with the active space of 12 orbitals and 12 electrons. Difference UV spectra for computationally described duplexes comprising the purinebromouracil stacks were also measured. We decided to compare the excited state properties of dimers in which the purine bases act as the 50 -neighbor of BrU, since, according to the literature data, such sequences differ substantially in their photoreactivity. Moreover, our calculations provided evidence in support of the hypothesis that electronic excitation leads to an ionic pair in which BrU is negatively and the 50 -neighboring purine base positively charged. The calculated energy of the first vertical (π,π*) transitions for the studied dimers stands in fair agreement with the maxima in the difference spectra measured for duplexes with the 50 -ABrU-30 or 50 -GBrU-30 sequences, which confirms the appropriate accuracy of our computational model. Furthermore, the MS-CASPT2 results show unequivocally that the charge transfer states, in which an electron is transferred from A/G to BrU, are located at much higher energies than the first (π,π*) transitions. In summary, the results of the current studies suggest the following sequence of processes leading to the formation of the reactive uracil-5-yl radical: (i) photons of 300 nm excite BrU in DNA to the first (π,π*) state of high oxidation potential (the CT states are not excited by these photons); (ii) if a DNA sequence enables electron transfer, the (π,π*) state oxidizes a distant guanine; 4536

dx.doi.org/10.1021/jp201028a |J. Phys. Chem. B 2011, 115, 4532–4537

The Journal of Physical Chemistry B (iii) if back electron transfer is sufficiently slow and the lifetime of the BrU radical anion sufficiently short, a bromide anion is released from BrU•
and the uracil-5-yl radical triggers chemical transformations leading to the experimentally observed damage.

’ ASSOCIATED CONTENT

bS

Supporting Information. Vertical excitation energies, oscillator strengths, and dipole moment differences in the y direction for the isolated adenine and guanine as well as the adenine-5-bromouracil and guanine-5-bromouracil dimers calculated at the CASSCF(12,12)/6-311G(d) and CASPT2/ CASSCF(12,12)/6-311G(d) levels of theory; the orbitals used to construct the active space for the studied dimers. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected].

’ ACKNOWLEDGMENT The work was carried out as part of the HPC-EUROPAþþ Project (project number 211437), with the support of the European Community
Research Infrastructure Action of the FP7 “Coordination and support action” Programme. This material is based on work supported by the Polish Ministry of Science and Higher Education (MNiSW), grant nos. N N204 023135 and DS/8221-4-0140-1, and the Spanish Ministry of Science and Innovation (MICINN), grant no. CTQ2008-06696. ’ REFERENCES (1) (a) Dewey, W. C.; Humphrey, R. M. Radiat. Res. 1965, 26, 538. (b) Dewey, W. C.; Sedita, B. A.; Humphrey, R. M. Science 1966, 152, 519. (c) Djordjevic, B.; Szybalski, W. J. Exp. Med. 1960, 112, 509 and references therein. (2) Cook, G. P.; Greenberg, M. M. J. Am. Chem. Soc. 1996, 118, 10025. (3) Cook, G. P.; Chen, T.; Koppisch, A. T.; Greenberg, M. M. Chem. Biol. 1999, 6, 451. (4) Chen, T.; Cook, G. P.; Koppisch, A. T.; Greenberg, M. M. J. Am. Chem. Soc. 2000, 122, 3861. (5) Hutchinson, F. Q. Rev. Biophys. 1973, 6, 201. (6) Krasin, F.; Hutchinson, F. Biophys. J. 1978, 24, 657. (7) Hewitt, R. R.; Marburger, K. Photochem. Photobiol. 1975, 21, 413. (8) Sugiyama, H.; Tsutsumi, Y.; Saito, I. J. Am. Chem. Soc. 1990, 112, 6720. (9) Zeng, Y.; Wang, Y. S. J. Am. Chem. Soc. 2004, 126, 6552. (10) Zeng, Y.; Wang, Y. S. Biochemistry 2007, 46, 8189. (11) Zeng, Y.; Wang, Y. Nucleic Acids Res. 2006, 126, 6521. (12) Cecchini, S.; Masson, C.; La Madeleine, C.; Huels, M. A.; Sanche, L.; Wagner, J. R.; Hunting, D. J. Biochemistry 2005, 44, 16957. (13) Zimbrick, J. D.; Ward, J. F.; Myers, L. S., Jr. Int. J. Radiat. Biol. 1969, 16, 525. (14) Pogozelski, W. K.; Tullius, T. D. Chem. Rev. 1998, 98, 1089. (15) Sugiyama, H.; Fujimoto, K.; Saito, I.; Kawashima, E.; Sekine, T.; Ishido, Y. Tetrahedron Lett. 1996, 37, 1805. (16) Wetmore, S. D.; Boyd, R. J.; Eriksson, L. A. Chem. Phys. Lett. 2000, 322, 129. (17) Fujimoto, K.; Sugiyama, H.; Saito, I. Tetrahedron Lett. 1998, 39, 2137. (18) Doddridge, Z. A.; Warner, J. L.; Cullis, P. M.; Jones, G. D. D. Chem. Commun. 1998, 1997.

ARTICLE

(19) Watanabe, T.; Tashiro, R.; Sugiyama, H. J. Am. Chem. Soc. 2007, 129, 8163. (20) Watanabe, T.; Toshikazu, B.; Xu, Y.; Tashiro, R.; Sugiyama, H. J. Am. Chem. Soc. 2005, 127, 44. (21) Tashiro, R.; Nakamura, K.; Sugiyama, H. Tetrahedron Lett. 2008, 49, 428. (22) Lu, X.-J.; Olson, W. K. Nucleic Acids Res. 2003, 31, 5108. (23) Clowney, L.; Jain, S. C.; Srinivasan, A. R.; Westbrook, J.; Olson, W. K.; Berman, H. M. J. Am. Chem. Soc. 1996, 118, 509. (24) Blancafort, L. J. Am. Chem. Soc. 2006, 128, 210. (25) Fleig, T.; Knecht, S.; H€attig, C. J. Phys. Chem. A 2007, 111, 5482. (26) Frisch, M. J. et al. Gaussian03, revision B.02; Gaussian Inc.: Pittsburgh, PA, 2003. (27) Karlstr€om, G.; Lindh, R.; Malmqvist, P.-Å.; Roos, B. O.; Ryde, U.; Veryazov, V.; Widmark, P.-O.; Cossi, M.; Schimmelpfennig, B.; Neogrady, P.; Seijo, L. Comput. Mater. Sci. 2003, 28, 222. (28) Ghigo, G.; Roos, B. O.; Malmqvist, P.-Å. Chem. Phys. Lett. 2004, 396, 142. (29) Roos, B. O.; Andersson, K.; F€ulscher, M. P.; Serrano-Andr
es, L.; Pierloot, K.; Merch
an, M.; Molina., V. J. Mol. Struct. (THEOCHEM) 1996, 388, 257. (30) Abouaf, R.; Pommier, J.; Dunet, H. Chem. Phys. Lett. 2003, 381, 486. (31) Kobyzecka, M.; Migani, A.; Asturiol, D.; Rak, J.; Blancafort, L. J. Phys. Chem. A 2009, 113, 5489. (32) Shukla, M. K.; Leszczynski, J. Chapter 6: Excited States of Nucleic Acid Bases. In Computational Chemistry: Reviews of Current Trends; Leszczynski, J., Ed.; World Scientific: Singapore, 2003; Vol. 8. (33) Crespo-Hern
andez, C. E.; Cohen, B.; Hare, P. M.; Kohler, B. Chem. Rev. 2004, 104, 1977. (34) Marian, C. M. J. Phys. Chem. A 2007, 111, 1545. (35) Serrano-Andr
es, L.; Merch
an, M.; Borin, A. C. J. Am. Chem. Soc. 2008, 130, 2473. (36) Bell, R. P. Trans. Faraday Soc. 1931, 27, 797. (37) Rak, J.; Voityuk, A. A.; Roesch, N. Int. J. Quantum Chem. 2000, 77, 128. (38) Li, X.; Sanche, L.; Sevilla, M. D. J. Phys. Chem. A 2002, 106, 11248. (39) Takaya, T.; Su, C.; de La Harpe, K.; Crespo-Hern
andez, C. E.; Kohler, B. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 10285.

4537

dx.doi.org/10.1021/jp201028a |J. Phys. Chem. B 2011, 115, 4532–4537